How-to
Pushing UDT definitions to Ignition Edge gateways with EAM
Short answer
Yes. An EAM Send Tags task pushes nested UDT definitions to every Edge gateway, because each gateway resolves its instances locally; a remote tag provider will not resolve a type held only on the main gateway. Expect remote tag delay of the 100 ms poll rate plus the network round trip, delivered over WebSockets.
On this page
- Can I push UDT definitions to my Edge gateways with EAM?
- Can an Edge gateway instantiate a UDT that exists only on the main gateway?
- How do I send the definitions with an EAM Send Tags task?
- How do I update definitions without overwriting site-specific settings?
- Can I edit tags on an Edge gateway from the main Designer?
- What delay should I expect when Edge is a remote tag provider?
- Do multiple register spans force a slower poll rate?
- Questions asked
Yes, EAM can push them. A Send Tags agent task transfers UDT definitions along with tags because tag types are path-addressed resources inside a tag provider, so a 40-gateway rollout becomes one repeatable task from the controller instead of 40 manual exports. What it cannot do is let an Edge gateway instantiate a type that exists only on the main gateway: each gateway resolves its instances against the definitions in its own providers, so the definitions must live on Edge. This walkthrough sends the nested definitions, keeps site-specific settings out of the blast radius, separates runtime writes from configuration edits, and then puts numbers on remote tag provider delay and Modbus poll pace.
Can I push UDT definitions to my Edge gateways with EAM? #
Yes: an EAM Send Tags task transfers UDT definitions along with tags, because tag types are resources with paths inside a tag provider, so one task from the controller replaces 40 manual export-and-import cycles for nested device types.
UDT definitions live under data types, and data types exist inside a tag provider — that is what gives a definition a path a task can select and send. Two limits hold. The task moves configuration, not values and not device connections. And nothing propagates a definition on its own: a remote tag provider carries values between gateways, never the type definitions a local instance needs to resolve.
Can an Edge gateway instantiate a UDT that exists only on the main gateway? #
No: a UDT instance resolves its type inside the tag provider where the instance is configured, so each of the 40 Edge gateways needs the nested definitions in its own provider before it can create device instances.
A remote tag provider exposes another gateway's tags — values, quality, timestamps — over the gateway network. It is a window, not a type store. When the main gateway browses an Edge instance through a remote provider, the instance still belongs to Edge: acquisition, type resolution, and local configuration stay there. The opposite design, where the main gateway owns the structures and Edge supplies values, is a different architecture covered by reusing Ignition edge tag structures on a central gateway. For your stated design — similar devices modeled locally on 40 Edge gateways — the definitions go to Edge, and this is what must exist on each one:
| Resource on each Edge gateway | Why it must be there | What to check |
|---|---|---|
| Nested member type definitions | The outer type references them by path | Every referenced type exists, with matching members and parameters |
| Outer device type definition | Defines the full instance layout | Members and parameters match the controller's release |
| Device connections | Bind instance members to device data | Connection names match the device references in the definitions |
| Instances and acquisition settings | Represent the devices at that site | Type reference resolves; addresses and rates carry site values |
How do I send the definitions with an EAM Send Tags task? #
Create the nested definitions once on the controller, select every referenced type path in a Send Tags agent task, send to one pilot Edge gateway, verify the instances resolve, then run the same task against the remaining 39 agents.
- Build the types on the controller: member types first, outer type last, then one instance against a real device connection, so unresolved references surface before anything is distributed.
- Create a Send Tags agent task on the controller and set the pilot Edge gateway as the destination agent.
- Select the type paths from the provider the Edge instances will use — every nested member type plus the outer type. An outer definition alone is not enough when its member types are missing or differ at the destination.
- Check the task's replacement behavior before running it: confirm what happens to tags and types that already exist at the destination, because that setting decides whether a re-send updates or overwrites.
- Confirm the EAM service security tag-editing permission covers the task. When it executes, the Send Tags task checks the service security tag-editing permissions rather than the destination provider's own editing permissions.
- Run the task against the pilot, then open the pilot's Designer, create one instance from the outer type, and confirm every nested member resolves and acquires from the local device connection.
- Apply the verified task to the remaining agents, sending into the same provider name on each gateway.
How do I update definitions without overwriting site-specific settings? #
Keep the shared structure in the definitions and the per-site differences in instance parameters, device references, and addresses, then check the task's replacement behavior before each send so a definition update does not discard local device configuration.
The structure lives in the type; the site lives in the instance. A device address, a poll rate, and a connection reference are instance-level values that survive a definition push which does not touch them. A member added to the type is a structural change and reaches every gateway you send to — which is what you want, and why the pilot run comes first. After any change, compare the member list and parameters on the pilot against the controller's intent before touching the other 39.
Can I edit tags on an Edge gateway from the main Designer? #
Runtime writes, yes; configuration edits, no: a remote tag provider forwards values and accepts value writes, but it does not open the owning gateway's tag configuration, so push definition and property changes to Edge with the EAM task instead.
Separate the two failures before you chase either. If reads through the main gateway also fail, you have a gateway connection problem. If reads succeed and only the edit is refused, that is the provider boundary working as designed, not a fault. Writing a runtime value to an Edge tag from the main gateway is supported; changing that tag's configuration is not, and the supported path for configuration change runs controller to agent.
What delay should I expect when Edge is a remote tag provider? #
About the poll rate plus the network round trip: remote tag providers deliver subscribed values over WebSockets, so a 100 ms Modbus poll on a local network reaches the main gateway about 102 ms after a value changes, worst case.
The WebSocket hop is not a poll — it pushes each change as Edge records it, so the interval that dominates end-to-end delay is the device poll rate. Worked example with illustrative but realistic numbers:
- Modbus device polled by Edge every 100 ms; switched LAN with 1 ms round trip to the device; gateway-to-gateway hop on the same network, 1 ms.
- A value changes just after a poll completes, so it waits for the next poll: worst case 100 ms, average 50 ms.
- That poll's response crosses the device network: +1 ms.
- Edge updates the tag and the subscription forwards it to the main gateway: +1 ms.
- Worst case: 100 + 1 + 1 = 102 ms. Average: 50 + 1 + 1 = 52 ms.
Raise the poll rate to 500 ms and the same arithmetic gives 502 ms worst case; the WebSocket delivery does not scale with it. The assumptions are one network hop and no congestion. Decision: values consumed at the main gateway on this scale are supervisory — trends, dashboards, history — while interlocking and fast control belong on the PLC or at Edge, never across a gateway hop. If you also need history from these tags, query tag history from a remote tag provider is its own setup path.
Do multiple register spans force a slower poll rate? #
No, as long as the response times of all requests fit inside the rate: contiguous spans coalesce into block reads of at most 125 registers each, and the pace cannot beat the sum of the per-request response times.
Your example — holding registers 0–99, 100–199, and 200–299 — is one contiguous block of 300 registers, so you do not divide your rate by three. Modbus limits a read request to 125 registers, so the driver issues a minimum of 3 requests per scan, and drivers optimize adjacent or nearly adjacent ranges into the largest blocks the device accepts, usually with an option to disable that optimization for devices that reject large requests. Worked example:
- 300 contiguous holding registers against the 125-register request limit: 3 requests per scan (125 + 125 + 50 = 300).
- Each request: 10 ms device processing + 1 ms network round trip = 11 ms.
- Scan time: 3 × 11 = 33 ms, which fits inside a 100 ms poll rate, so configure 100 ms and it holds.
- The same device answering in 40 ms instead: 3 × 41 = 123 ms per scan, so the effective pace settles near 123 ms regardless of the configured rate.
The 125-register limit is the protocol; the 10 ms and 40 ms response times are example inputs, so substitute the numbers from your driver's diagnostics page, which shows the request count and response times for all subscribed items. Those two numbers decide the pace on your hardware. Where a device documents a maximum poll rate, that is the ceiling; where it does not, the measured scan time is. Some controllers service communications in leftover scan time, so trust the measured response over the configured rate.
Questions asked
Does the Send Tags task also move the Modbus or OPC device connections?
No. Device connections are separate gateway resources, so create them on each Edge gateway yourself and give them the same names the definitions reference, so instance members bind on arrival. Configure and prove the connections once on a pilot gateway, then copy those settings to the rest before you send tags.
How do I confirm the definitions arrived on every Edge gateway?
EAM creates system tags on the controller for each agent, so build a monitoring view or alarms on those tags to see task completion across the fleet. On the gateway itself, create one test instance from the outer type in the Edge Designer and confirm every nested member resolves with the right parameters before you trust the rollout.
Two sites need slightly different behavior from the same device type. One definition or two?
One definition with parameters, as long as the member layout is identical. Put the differences in instance parameter values and device references, and the shared send keeps working. Split into two types only when the structure itself differs, because a member that exists on one site but not the other is a structural change, not a parameter.
Is a manual export and import good enough instead of EAM?
It works once and decays after that. Exporting tag types from the controller and importing them on 40 Edge gateways is 40 hand operations per change, with no status tracking and no repeatable ordering. The Send Tags task performs the same transfer on demand, reports through the controller, and stays identical across every update.